Kubernetes
Prerequisites
-
A Kubernetes cluster configured with kubectl
-
Helm installed locally
Resource requirements
This chart comes with three default deployment sizes: small, medium, and large
You can specify any upon installation using for example --set size=large
You can also override any individual request using --set onsightRequests.<resource>=<value>
where <resource> is cpu, memory, or ephemeral
Size |
Suggested # of Pods |
Requested OnSight CPUs |
Requested OnSight Memory |
Requested OnSight Ephemeral Storage |
---|---|---|---|---|
small |
< 500 |
1.0 |
1Gi |
10Gi |
medium |
500-2000 |
2.0 |
2Gi |
20Gi |
large |
>2000 |
3.0 |
3Gi |
50Gi |
Deploying FortiMonitor
See above for determining size. The default is medium.
If your cluster already has metrics-server installed, disable it as described below.
-
Add this Helm repo using
helm repo add panopta https://panopta.github.io/kubernetes/repo
-
Install FortiMonitor using
helm install --set customer_key=YOUR-CUSTOMER-KEY <name-of-release> panopta/panopta
In a few minutes, your cluster should show up in the FortiMonitor control panel.
A Note on metrics-server
The chart will install metrics-server by default. If you already have metrics-server installed in your cluster, you can skip it with --set metricsServer.install=false
Advanced configuration
If you wish to further customize your FortiMonitor deployment, you can pass additional options to the install command by adding one-to-many--set <key>=<value>
to the install command.
Available configuration options are listed in the following table.
You can also specify such options in a YAML-formatted values.yaml file which you can then pass along to the install command with -f values.yaml
Configuration Options
Key Name |
Default |
Description |
---|---|---|
customer_key |
None (Required) |
Your FortiMonitor customer key |
size |
medium |
Size of the cluster you are deploying to |
clusterName |
Kubernetes Cluster |
The name of this cluster as it will show up in the control panel |
metricsServer.install |
true |
Whether to install metrics-server as part of the deployment. Set to false if it’s already installed. |
topNNamespaces |
0 |
The number of namespaces to pull in, ordered by the number of pods. 0 to include all. |
onsightRequests.cpu |
3.0 |
Requested CPU for the FortiMonitor OnSight |
onsightRequests.memory |
3Gi |
Requested Memory for the FortiMonitor OnSight |
onsightRequests.ephemeral |
None |
Requested Ephemeral Storage for the FortiMonitor OnSight |
agent_config |
None |
Any additional blocks of configuration to deploy onto the nodes’ agents |
Upgrading FortiMonitor
-
Fetch the new chart using
helm repo update
-
Upgrade your deployment using
helm upgrade <deployment name> panopta/panopta
Uninstalling FortiMonitor
Run helm uninstall <release_name>
You can find the name of the release with helm ls